JP6276102B2 - Image processing apparatus, radiation tomography apparatus, and program - Google Patents

Description

  The present invention relates to a technique for noise reduction in an image obtained by radiation tomography.

  Conventionally, noise reduction processing has been performed on images obtained by radiation tomography. As noise reduction processing, in particular, a technique using smoothing processing is widely used (see Patent Document 1, paragraph 0034, etc.).

  According to such a method, it is possible to relatively easily reduce quantum noise of radiation that appears in a granular form in an image obtained by radiation tomography.

JP2012-070814A

  However, noise reduction processing using conventional smoothing processing often suppresses not only image noise but also images of a minute structure of a subject, for example, a minute blood vessel or bone.

  Under such circumstances, there is a demand for a technique capable of suppressing image noise while maintaining an image representing a minute structure of a subject in an image obtained by radiation tomography.

The invention of the first aspect
In an image obtained by radiological tomography of a subject, extraction means for extracting a radiation high absorption region by threshold processing of pixel values;
Detecting means for detecting a region having a predetermined level (level) or less of the extracted radiation high absorption region as a region of the microstructure,
A first noise reduction is performed for a non-microstructure region that is different from the microstructure region in the entire region of the image, and the first noise reduction is performed for the microstructure region. There is provided an image processing apparatus including processing means for performing a predetermined process in which a second noise reduction having a noise reduction effect smaller than the noise reduction is performed or a noise reduction is not substantially performed.

The invention of the second aspect is
The detection means is
A region of interest of a predetermined size is set in the plane direction for each position on the image, and the number of pixels in the radiation high absorption region existing discontinuously outside the region of interest within the region of interest is counted. The image processing apparatus according to the first aspect is provided, wherein at least the number of pixels is equal to or less than a predetermined value, and the radiation high absorption region is detected as a region of a minute structure.

The invention of the third aspect is
The image is a three-dimensional image;
The detection means has a number of pixels in the radiation high absorption region that are discontinuously present outside the region of interest within the region of interest set on the predetermined two-dimensional image included in the three-dimensional image. And at least a part of the radiation superabsorbing region in the region of interest passes continuously through all the plurality of continuous two-dimensional images including the predetermined two-dimensional image in the three-dimensional image. The image processing according to the second aspect, wherein the radiation high absorption region in the region of interest on the predetermined two-dimensional image is detected as a region of the microstructure when included in the radiation high absorption region to be connected. Providing equipment.

The invention of the fourth aspect is
Any one of the first to third aspects, wherein the extraction unit sets a threshold value of the threshold value processing based on a noise estimation amount of a pixel value and / or a histogram in at least a part of the image. An image processing apparatus according to one aspect is provided.

The invention of the fifth aspect is
The image processing apparatus according to the fourth aspect, in which the extraction unit sets the threshold based on a noise estimation amount of the image.

The invention of the sixth aspect is
The predetermined process is a weighted addition process between the image and an image obtained by performing noise reduction processing on the image, and the weighting used for the non-microstructure area and the microstructure area There is provided an image processing apparatus according to any one of the first to fifth aspects, wherein the weighting process is different from each other.

The invention of the seventh aspect
The predetermined process is a noise reduction process in which a noise reduction effect changes according to a parameter, and the parameter used for the non-microstructure area and the microstructure area There is provided an image processing apparatus according to any one of the first to fifth aspects, wherein the parameters used are noise reduction processes different from each other.

The invention of the eighth aspect
An image processing apparatus according to any one of the first to seventh aspects, wherein the microstructure is a blood vessel into which a contrast medium has been injected.

The invention of the ninth aspect is
The image processing apparatus according to any one of the first to seventh aspects, wherein the microstructure is a bone part.

The invention of the tenth aspect is
A radiation tomography apparatus comprising the image processing apparatus according to any one of the first to ninth aspects is provided.

The invention of the eleventh aspect is
A program for causing a computer to function as the image processing apparatus according to any one of the first to ninth aspects is provided.

  According to the invention of the above aspect, in the image obtained by radiation tomography, the microstructure is detected based on the feature relating to the form of the microstructure on the image, and the region of the microstructure is non-microstructure. Since the image noise reduction processing is performed so that the noise reduction effect is smaller than the object region, it is possible to accurately detect the minute structure and reduce the noise while maintaining the image.

1 is a diagram schematically showing a configuration of an X-ray CT apparatus (X-ray computed tomography system) according to the present embodiment. It is a functional block (block) figure which shows the structure of the part in connection with the image noise reduction process in a X-ray CT apparatus. It is a flow figure of the noise reduction process by this embodiment. It is a figure which shows the typical example of the histogram of CT value of each pixel in the medical image of the biological body which inject | poured the contrast agent to the blood vessel. It is a figure which shows the example of a setting of an attention area. It is a figure which shows the example of an attention tomogram group. It is a figure for demonstrating an example of the process by a noise reduction process part.

  Embodiments of the invention will be described below. However, this does not limit the invention.

  FIG. 1 is a diagram schematically showing a configuration of an X-ray CT apparatus according to the present embodiment.

  As shown in FIG. 1, the X-ray CT apparatus 100 includes an operation console 1, an imaging table 10, and a scanning gantry 20.

  The operation console 1 includes an input device 2 that receives input from the operator 41, data for performing control of each unit for imaging the subject (subject) 40, data processing for generating an image, and the like. A processing device 3, a data collection buffer (buffer) 5 that collects data acquired by the scanning gantry 20, a monitor 6 that displays an image, and a storage device 7 that stores programs, data, and the like are provided. .

  The imaging table 10 includes a cradle 12 on which the subject 40 is placed and conveyed to the cavity B of the scanning gantry 20. The cradle 12 is moved up and down and horizontally moved by a motor built in the imaging table 10. Here, the body axis direction of the subject 40, that is, the horizontal linear movement direction of the cradle 12 is the z-axis direction, the vertical direction is the y-axis direction, and the horizontal direction perpendicular to the z-axis direction and the y-axis direction is the x-axis direction. .

  The scanning gantry 20 includes a rotating unit 15 that is rotatably supported. The rotating unit 15 includes an X-ray tube 21, an X-ray controller 22 that controls the X-ray tube 21, and the X-ray 81 generated from the X-ray tube 21 is shaped into a fan beam or a cone beam. Aperture 23, X-ray detector 24 that detects X-ray 81 transmitted through the subject 40, DAS 25 that collects output signals of the X-ray detector 24 as data, X-ray controller 22, aperture 23 And a rotating unit controller 26 for controlling the above. The main body of the scanning gantry 20 includes a control controller 29 that communicates control signals and the like with the operation console 1 and the imaging table 10. The rotating unit 15 and the main body of the scanning gantry 20 are electrically connected via a slip ring 30.

  The X-ray tube 21 and the X-ray detector 24 are arranged to face each other with the imaging space in which the subject 40 is placed, that is, the cavity B of the scanning gantry 20 interposed therebetween. When the rotating unit 15 rotates, the X-ray tube 21 and the X-ray detector 24 rotate around the subject 40 while maintaining the positional relationship. A fan beam or cone beam X-ray 81 emitted from the X-ray tube 21 and shaped by the aperture 23 passes through the subject 40 and is irradiated onto the detection surface of the X-ray detector 24.

  Here, the spreading direction of the X-ray 81 of the fan beam or cone beam in the xy plane is the channel direction (CH direction), the spreading direction in the z-axis direction or the z-axis direction itself is the slice direction (SL direction), A direction toward the rotation center of the rotation unit 15 in the xy plane is represented by an isocenter direction (I direction).

  The X-ray detector 24 includes a plurality of detection elements 24i arranged in the channel direction and the slice direction. The number of detection elements 24i in the channel direction is, for example, about 1000 in the angle range of 60 °, and the arrangement interval is, for example, about 1 mm.

  Thus, the image noise reduction processing according to the present embodiment will be described.

  FIG. 2 is a functional block diagram showing a configuration of a part related to image noise reduction processing in the X-ray CT apparatus according to the present embodiment. As shown in FIG. 2, the X-ray CT apparatus 100 includes an X-ray high absorption region extraction unit 31, a minute structure detection unit 32, and a noise reduction processing unit 33. These units are functionally realized by the data processing device 3 reading and executing a program stored in the storage device 7.

  FIG. 3 is a flowchart of image noise reduction processing according to the present embodiment.

  Note that a plurality of tomographic images are already stored in the storage device 7. In this example, the plurality of tomographic images are obtained by X-ray CT imaging of a subject in which a contrast medium is injected into a blood vessel. Each tomographic image is a slice image having a cross section along the xy plane. The image size of the tomographic image is 512 × 512 pixels. The plurality of tomographic images are continuously arranged in the z direction and constitute a three-dimensional image.

  In step S <b> 1, the X-ray high absorption region extraction unit 31 reads and acquires the three-dimensional image V of the subject stored in the storage device 7.

  In step S2, the X-ray high absorption region extraction unit 31 estimates the amount of image noise in the three-dimensional image V acquired in step S1. In this example, particulate noise is assumed as noise, and the image SD value is obtained as the amount of image noise. The image SD value is, for example, the standard deviation of the CT value in a region where an essentially uniform CT value is assumed in the three-dimensional image V.

  In step S3, the X-ray high absorption region extraction unit 31 obtains a CT value corresponding to the boundary between the soft tissue and the contrasted blood vessel in consideration of the estimated image noise amount, and uses the CT value for threshold processing in the next step. Set as the threshold to be used.

  Here, a threshold value setting method will be described.

  FIG. 4 shows a typical example of a histogram of CT values of each pixel in a medical image of a living body in which a contrast medium is injected into a blood vessel. In this histogram, the horizontal axis represents the CT value and the vertical axis represents the frequency. A living body is mainly composed of a soft tissue, a contrasted blood vessel, and a bone. In general, CT values corresponding to soft tissues take a positive low value, and CT values corresponding to contrasted blood vessels and bones take a positive high value. Therefore, in the histogram of FIG. 4, a mountain with a positive CT value close to a low value mainly corresponds to a soft tissue, and a mountain with a positive CT value close to a high value mainly corresponds to a contrasted blood vessel and a bone part. Therefore, the CT value corresponding to the boundary between the soft tissue, the contrasted blood vessel, and the bone can be a CT value corresponding to the valley between these two peaks. However, the actual boundary is affected by the noise of the image, and the CT value is shifted to the positive side on the average by the amount of noise. Therefore, in this example, a typical CT value CB corresponding to the boundary between the soft tissue and the contrasted blood vessel and the bone in an ideal image without noise and a typical CT value ST of the soft tissue depend on the noise amount SD. The value ST + α · SD formed by adding the elements α · SD to be compared is compared, and the larger one is set as the threshold value T1. The value ST can be 60 to 80, the value CB can be 90 to 110, and the value α can be 0 to 1, for example. From experience, it is appropriate that the value ST is, for example, 70, the value CB is, for example, 100, and the value α is, for example, about 0.5.

  In step S4, the X-ray high absorption region extraction unit 31 determines, for each pixel constituting the three-dimensional image, whether or not the CT value of the pixel is equal to or greater than the threshold value T1. If it is equal to or greater than the threshold value T1, the pixel is extracted as an X-ray high absorption region C0 including a contrasted blood vessel and a bone part. It should be noted that a pixel having a very high CT value that is not less than a certain value may not be extracted as the X-ray high absorption region C0 so that noise other than contrast blood vessels and bones is not erroneously extracted.

  In step S <b> 5, the microstructure detection unit 32 selects, as the target tomographic image A, an arbitrary one-slice tomographic image having the xy plane as a cross section.

  In step S <b> 6, the microstructure detection unit 32 sets a region of interest F having a predetermined size at an arbitrary position on the target tomographic image A. The attention area F can be a rectangular area of 5 × 5 pixels or 7 × 7 pixels, for example. In this example, the attention area F is a rectangular area of 5 × 5 pixels.

  FIG. 5 shows an example of setting the attention area. In the example of FIG. 5, there are four X-ray high absorption region pixels in the attention area F of 5 × 5 pixels.

  In step S7, the minute structure detection unit 32 counts the number of pixels in the X-ray high absorption region existing in the region of interest. In the example of FIG. 5, four are counted. Then, the microstructure detection unit 32 determines whether or not the counted number N of the first candidate pixels C1 is equal to or less than the predetermined number T2. When the counted number N of pixels is equal to or less than the predetermined number T2, the X-ray high absorption region C0 in the region of interest F has a surface direction spread below a predetermined level, and is a contrast blood vessel or a microstructure of a bone part. It is determined that there is a possibility, and is extracted as the first candidate pixel C1 of the minute structure. Then, the process proceeds to step S9. On the other hand, when the counted number N of pixels exceeds the predetermined number T2, the X-ray high absorption region C0 in the region of interest F does not have a predetermined level or less, and is less likely to be a microstructure. It progresses to step S11. The predetermined number T2 can be set to about 3 to 7, for example. In this example, the predetermined number T2 is 4. In the example of FIG. 5, since the number of pixels of the X-ray high absorption region is 4, these pixels are extracted as the primary candidate pixels C1.

  In step S <b> 8, the minute structure detection unit 32 determines whether or not the primary candidate pixel C <b> 1 in the attention area F is discontinuous inside and outside the attention area F. For example, when the attention area F is an area of 5 × 5 pixels, the primary candidate pixel C1 in the attention area F and the first candidate pixel C1 in the peripheral area of 7 × 7 pixels surrounding the area The determination is made based on whether or not there are candidate pixels that are continuous in an oblique direction. When all the primary candidate pixels C1 are discontinuous inside and outside the attention area F, it is determined that the first candidate pixel C2 in the attention area F is more likely to be a microstructure, and the microstructure Extracted as the second candidate pixel C2 of the object. Then, the process proceeds to step S9. On the other hand, when at least a part of the primary candidate pixels is discontinuous inside and outside the attention area F, it is determined that the first candidate pixel C1 in the attention area F is unlikely to be a microstructure. The process proceeds to step S11. In the example of FIG. 5, since these are discontinuous inside and outside the attention area F, these pixels are extracted as secondary candidate pixels C2.

  In step S9, the microstructure detection unit 32 determines whether at least a part of the secondary candidate pixels C2 in the attention area F is continuous along the direction of a straight line or a curve that is not parallel to the xy plane. judge. For example, a plurality of tomographic images including the target tomographic image A and continuously arranged in the z direction are designated as the tomographic image group Q of interest. This tomographic image group Q can be, for example, three or five tomographic images that are continuous in the z direction with the target tomographic image A as the center. In this example, three tomographic images continuous in the z direction are used. Next, in the tomographic image group Q, the microstructure detection unit 32 connects the tomographic images adjacent in the z direction within ± 1, that is, within 3 × 3 9 pixels in the xy plane. The continuous pixel group of the X-ray high absorption area | region connected by is specified. Then, it is determined whether or not any of the secondary candidate pixels C2 in the attention area F in the target tomographic image A is included. If it is included, it is determined that at least a part of the second candidate pixel C2 in the attention area F is continuous along a direction non-parallel to the xy plane. In this case, it is determined that the second candidate pixel C2 in the attention area F is not a non-continuous noise in a direction non-parallel to the xy plane, but is a minute structure, and the process proceeds to step S10. On the other hand, when all of the secondary candidate pixels C2 in the attention area F are discontinuous along the direction non-parallel to the xy plane, the second candidate pixels C2 in the attention area F are not in the xy plane. It is determined that the noise is not continuity in the non-parallel direction and is not a minute structure, and the process proceeds to step S11. In the tomographic image group Q of interest, when specifying a continuous pixel group in the X-ray high absorption region where the xy coordinates are within ± 1 for each tomographic image adjacent in the z direction, Among them, the discontinuity inside and outside the attention area F of the pixel of the X-ray high absorption area in the tomographic image other than the target tomographic image A may be included in the specific requirement, but may not be included when the calculation amount is suppressed. .

  In step S10, the second candidate pixel C2 in the attention area F is extracted as the area of the minute structure.

  FIG. 6 shows an example of the tomographic image group Q of interest. In the example of FIG. 6, since a part of the secondary candidate pixels C2 constitutes a pixel group that continues in the z direction, all these secondary candidate pixels C2 are detected as a region of a minute structure.

  In step S11, the minute structure detection unit 32 determines whether there is a position where a new attention area should be set. If there is, the process returns to step S6 to change the position and set a new attention area.

  In step S12, the microstructure detection unit 32 determines whether there is a tomographic image to be set as a new target tomographic image. If there is, the process returns to step S5, and a new target tomographic image is selected by changing the slice position.

  In step S13, the noise reduction processing unit 33 performs predetermined processing for noise reduction on the three-dimensional image. Specifically, the first noise reduction is performed for the non-microstructure region different from the microstructural region in the entire region of the three-dimensional image, and for the microstructural region. A predetermined process in which the second noise reduction, which has a smaller noise reduction effect than the first noise reduction, is performed or the noise reduction is not substantially performed is performed.

  For example, the predetermined processing is weighted addition processing of the original three-dimensional image and the processed three-dimensional image subjected to noise reduction processing, and the weight of the processed three-dimensional image is applied to the non-microstructure region. The weight of the original three-dimensional image is increased for the region of the microstructure.

  FIG. 7 shows that in this processing, in particular, the weight of the processed 3D image for the non-microstructure region is 1 (the weight of the original 3D image is 0), and the original 3D image is for the microstructure region. The processing when the weight of 1 is 1 (the weight of the processed three-dimensional image is 0) is shown. In other words, in this example, noise reduction processing is performed on the subtracted minute structure extracted from the original three-dimensional image. Then, an image subjected to noise reduction processing is adopted for the area other than the minute structure, and the original three-dimensional image is adopted for the area of the minute structure, which are added together and synthesized.

  Further, for example, in the predetermined processing, noise reduction processing in which the noise reduction effect changes according to the parameters is performed on the three-dimensional image, and the noise reduction effect is enhanced for the non-microstructure region. The parameter is set so that the noise reduction effect is low for the region of the minute structure.

  As the noise reduction process, for example, a smoothed image filter process using a mask such as 3 × 3 pixels or 5 × 5 pixels can be considered. Alternatively, in the image space or the projection data space, for each local region, based on the “noise amount” based on the size or variation of the data value and the correlation of the data value between the local region and the adjacent region of the local region It can be considered that a “noise reduction weight” is obtained and a noise suppression process is performed using these “noise amount” and “noise reduction weight” as parameters.

  As described above, according to the present embodiment, the minute structures such as contrast blood vessels and bones have a large CT value, do not spread in the surface direction, and have continuity in a direction non-parallel to the surface. Based on the detection, image noise reduction processing is performed so that the noise reduction effect is smaller in the area of the microstructure than in the area of the non-microstructure, so that the microstructure is accurately detected and the image is maintained. However, noise can be reduced.

  In this embodiment, when determining whether or not the extracted X-ray high absorption region pixel is a microstructure, there is no continuity along the direction of a straight line or a curve that is not parallel to the xy plane. It is determined that it is not a microstructure. As a result, the chance of erroneously detecting the non-continuous noise or the like as a microstructure can be reduced, and a highly accurate microstructure can be detected.

  The invention is not limited to this embodiment, and various modifications can be made without departing from the spirit of the invention.

  For example, in the present embodiment, a three-dimensional image including a plurality of continuous tomographic images is assumed as a processing target, but a two-dimensional image such as a single tomographic image can also be assumed. In this case, determination processing based on continuity along a direction non-parallel to the image plane of the image is not applied.

  Further, for example, in the present embodiment, the continuity of the X-ray high absorption region in the cross-sectional direction of the target tomographic image A is not determined, but only those having a certain degree of continuity are extracted as candidates for the minute structure. You may do it.

  Further, for example, in the present embodiment, the attention area F is a rectangular area, but it may be a circular area or the like.

  Further, for example, in the present embodiment, a contrasted blood vessel and a bone part are assumed as the minute structure. However, in the case of a subject who does not inject a contrast medium, a bone part can be assumed.

  For example, although this embodiment is an X-ray CT apparatus, the image processing apparatus which performs said image processing is also an example of embodiment of invention. A program for causing a computer to function as such an image processing apparatus, a storage medium storing the program, and the like are also examples of embodiments of the invention.

  In addition, for example, the present embodiment is an X-ray CT apparatus, but the invention can also be applied to a PET-CT apparatus, a SPECT-CT apparatus, or the like that combines an X-ray CT apparatus and PET or SPECT.

1 Operation Console 2 Input Device 3 Data Processing Device 5 Data Collection Buffer 6 Monitor 7 Storage Device 10 Imaging Table 12 Cradle 15 Rotating Unit 20 Scanning Gantry 21 X-ray Tube 22 X-ray Controller 23 Aperture 24 X-ray Detector 25 Detector Controller 26 Rotation unit controller 28 X-ray detection device 29 Control controller 30 Slip ring 31 X-ray high absorption region extraction unit 32 Microstructure detection unit 33 Noise reduction processing unit 40 Subject 41 Operator 81 X-ray 100 X-ray CT device

Claims (11)

In an image obtained by radiological tomography of a subject, extraction means for extracting a radiation high absorption region by threshold processing of pixel values;
Detecting means for detecting a region having a predetermined extent or less in the plane direction of the extracted radiation-absorbing region as a region of a microstructure;
A first noise reduction is performed for a non-microstructure region that is different from the microstructure region in the entire region of the image, and the first noise reduction is performed for the microstructure region. An image processing apparatus comprising: a processing unit that performs a predetermined process in which a second noise reduction having a noise reduction effect smaller than the noise reduction is performed or a noise reduction is not substantially performed. The detection means includes
A region of interest having a predetermined size is set for each position on the image, and the number of pixels in the radiation high absorption region existing discontinuously outside the region of interest in the region of interest is counted. The image processing apparatus according to claim 1, wherein when the number is equal to or less than a predetermined value, the radiation high absorption region is detected as a region of a minute structure. The image is a three-dimensional image;
In the detection unit, the number of pixels of the radiation high absorption region existing discontinuously outside the region of interest within the region of interest set on the predetermined two-dimensional image included in the three-dimensional image is equal to or less than a predetermined value. And at least a part of the radiation superabsorbing region in the region of interest passes continuously through all the plurality of continuous two-dimensional images including the predetermined two-dimensional image in the three-dimensional image. The image processing according to claim 2, wherein when included in the radiation high-absorption region that is connected, the radiation high-absorption region in the region of interest on the predetermined two-dimensional image is detected as a region of the microstructure. apparatus.   The image processing according to any one of claims 1 to 3, wherein the extraction unit sets a threshold value of the threshold value processing based on a noise estimation amount of a pixel value and / or a histogram in at least a part of the image. apparatus.   The image processing apparatus according to claim 4, wherein the extraction unit sets the threshold based on a noise estimation amount of the image.   The predetermined process is a weighted addition process of the image and an image obtained by performing noise reduction processing on the image, and the weighting used for the non-microstructure area and the microstructure area The image processing apparatus according to any one of claims 1 to 5, wherein the weighting processing is different from each other in weighting used.   The predetermined process is a noise reduction process in which a noise reduction effect changes according to a parameter, and is used for the parameter used for the non-microstructure region and the microstructure region. The image processing apparatus according to claim 1, wherein the parameters are noise reduction processes that differ from each other.   The image processing apparatus according to claim 1, wherein the microstructure is a blood vessel into which a contrast medium is injected.   The image processing apparatus according to claim 1, wherein the microstructure is a bone part.   A radiation tomography apparatus comprising the image processing apparatus according to claim 1.   A program for causing a computer to function as the image processing apparatus according to any one of claims 1 to 9.